Skip to main content

Full text of "The attitude-behavior relationship"

See other formats



no- 110 

Oz,f- A 

Faculty Working Papers 

ersity of Illinois at Urbana-Champaign 

Michael Ross 
University of Waterloo 

May 1, 1973 

College of Commerce and Business Administration 

University of Illinois at Urbano-Champaign 

The Attitude-Behavior Relationship 

Bobby J. Calder 
University of Illinois at Urbana-Champaign 

Michael Ross 
University of Waterloo 

May 1, 1973 


Attitude-Behavior Relationship 

Please send proofs to Bobby J, Calder, Organizational Behavior Program, 
194 Commerce West, University of Illinois, Urbana, Illinois 61801 


The purpose of this review is to establish the current status of 
research relating attitudes and behavior. It is argued that special 
attention must be given to theoretical conceptualizations of attitude 
and behavior. When a representative sample of studies from a number 
of areas are drawn together, it is possible to specify conditions 
under which attitudes and behavior are related. Explanations for 
failures to demonstrate this relationship are emphasized in order to 
direct future research. 



The Attitude-Behavior Relationship 
May 1, 1973 

The failure to demonstrate an unequivocal relationship between 
attitudes and behavior has been one of the more persistent problems in 
the social sciences. Since LaPlere (1934) reported an inconsistency 
between an indicant of racial attitude and behavior toward Chinese, 
numerous investigators have concluded that attitudes seem to be unre- 
lated to behavior (cf. Deutscher, 1969; Wicker, 1969). At the same 
time, attitude, to use Allport's oft-cited phrase, has been "the 
primary building stone in the edifice of social psychology," and has 
served as a major tool in formulating social and business policy. 
Unless the relationship between attitudes and behavior can be estab- 
lished, however, it is not at all clear that this reliance on the con- 
cept of attitude has been justified. In fact, the negative findings 
in this area have contributed to the rejection by some of the entire 
concept of attitude, Bern (1965, 1967, 1972), for example, has 
theorized that people do not have attitudes but rather infer them from 
their behavior. Similar thinking by Mischel (1968) with regard to 
personality traits raises the entire question of the need for postu- 
lating covert internal states. Such views have gained increased 
currency with the publication of Skinner *s Beyond Freedom and Dignity , 
The policy implications of this debate are enormous: Should social 
programs attempt to change people or should they concentrate on the 
environment in which people live (cf. Etzionni, 1972)? 

It is not yet certain that the pessimism concerning attitude is 
Justified. Although there have been a few excellent reviews of 

Digitized by tlie Internet Archive 

in 2011 witli funding from 

University of Illinois Urbana-Champaign 

attitude-behavior studies (e.g., Festinger, 1964; Wicker, 1969), they 
have generally addressed rather specific interests or have been exposi- 
tory in nature. There has been little effort to review and draw 
together all of the disparate findings relevant to the attitude- 
behavior relationship. Not even textbook accounts such as the one 
by Keisler, Collins and Miller (1968), though valuable, make any 
pretense of being integrative. The purpose of this paper is to 
review a truly representative sample of studies and to delineate the 
current status of this research in specifying the relationship between 
attitudes and behavior. 

We will begin by discussing briefly the conceptual nature of both 
attitude and behavior. Much of the attitude literature has been able 
to bypass such considerations because of its focus on attitude change, 
since any change in a measure of attitude may be assumed to reflect 
some change in the unspecified concept of attitude. In contrast, the 
conceptual underpinnings of attitude cannot be so easily ignored if 
attitudes are to be related to behavior. 

Theoretical Conceptualization 


As McGuire (1968) has noted, conceptions of attitude vary along a 
continuum from positivistic to latent process. Positivistic research- 
ers equate attitudes with the operations designed to measure them 
(e.g., DcFleur and Westie, 1963). Attitude may thus be defined 
entirely in terms of a person's responses to an attitude question- 
naire just as I.Q. may be equated with a score on an I.Q. test. 

Despite its appealing simplicity, the positivistic approach is prob- 
lematic. If a theoretical explanatory structure approaches the com- 
plexity of natural events, then W' can no more understand our theory 
than we can the real world V7e seek to explain. Investigators at the 
other end of the continuum view attitude as a latent process. Some 
propose that this latent process is an inference based on response 
consistency (e,g,, Campbell, 1959) while others assume that it is 
literally an affective or drive-like state within the organism (Doob, 
19A7; Thurstone, 1931). Attitude is probably best conceived, however, 
as a theoretical abstraction useful for explanation and prediction. 
This conception of attitude does not imply any necessary behavioral 
or phenomeno logical referent. 

At the level of measurement, behavior in the form of self-reports, 
physiological reactions, and other overt responses serve as indicants 
of attitude. If a behavior is to be accepted as an indicant of atti- 
tude, it must constitute an appropriate operationalization of the 
Concept, That is, there must exist some theoretical rationale for 
inferring an attitude from that behavior. There can be no inconsis- 
tency between an attitude and its behavioral indicant, otherwise we 
would have no knowledge of the attitude. This does not imply, as is 
sometimes concluded (e,g, , Sechrest, 1969, p. 147), that attitude- 
behavior inconsistency cannot exist by definition-- an attitude may 
be inconsistent with other behaviors that are not direct dperationa- 
iizations of the attitude. 

No doubt much of the confusion centering around this point lies 
in the difficulty of operational izing attitude. Though several 



classes of attitude measures may be distinguished (Cook and Selltiz, 
1964), only the traditional attitude scaling techniques are satisfac- 
tory as operationalizations. The implicit, and in fact ill- 
acknowledged, rationale for attitude scaling is that attitudes are 
derived from the informational beliefs one has about an issue (cf. 
Ostrom, 1968; Calder, Insko, and Yandell, in press). Unfortunately, 
the other classes of measures have little rationale at all. Obser- 
vational techniques (e.g., Webb, Campbell, Schwartz and Sechrest, 
1965) while often intriguing seem particularly weak ii. this respect. 
For example, in one study, investigators chose seating aggregation as 
an Index of attitude (Campbell, Kruskal and Wallace, 1966). Observers 
noted the extent to which black and white studett s tended to sit apart 
in lecture halls at two universities. Racial separation was found to 
be greater at one of the schools. As a result the authors suggested 
that the white students at this school v/ere more prejudiced than the 
white students at the other school. Is oeating aggregation, however, 
Operationallzation of attitude, o?- ±s it more appropriately viewed as 
& behavior that under some cJ^rcumstances might be under the control of 
(or predicted from) attitudes? In short, there is often no theoretical 
rationale for taking an observatior of overt behavior as an operational* 
ization of an attitude. In terms of the general attitude-behavior 
relationship, attitude must first be adequately operationalized and 
measured if it is to be related to overt behaviors. 

As the above discussion points up, behavior serves a dual role 
in psychology. First it represents what most psychologists wish to 
understand or predict. Behavior Is the object of psychological 

theory. In order to understand behavior, concepts such as attitude 
are developed. Yet, in deriving these concepts, behavior must serve a 
second role. Behavior must function as the evidence by which these 
concepts are inferred and quantified. Behavior in some form is the 
only source of data about psychological processes. We are thus faced 
with the problem of using behavior to develop concepts which will aid 
in understanding behavior. All of this has fundamental implications 
for the attitude-behavior relationship. Behavior must serve as the 
evidence for attitudes. At the same time, to be of any value an 
attitude must aid in understanding and predicting behavior that is not 
a direct operationalization of that attitude. 


What is it that attitudes should be behavioral ly related to? 
Much confusion concerning the attitude-behavior relationship has 
centered around this question. Most investigators have not treated 
behavior as an abstract concept but have attempted instead to relate 
attitudes to quite specific acts. For example, LaPiere's study inves- 
tigated whether an attractive Chinese couple was refused service at 
hotels and restaurants. There is a need for a broader theoretical 
treatment of behavior. Attitudes should be related to rather general 
behavioral syndromes or tendencies to perform a class of actions 
rather than to single acts. Any one instarscc of an act is neces- 
sarily too overdetermined to test the attitude-behavior relationship. 
Although most studies have followed LaPiere in using a single act as 
a measure of behavior, Fishbein (in press) has recently approached 

this problem by viewing behavioral observations as criterion scores. 
He argues that ideally such behavior scores should be based on repeated 
observations of multiple acts. While studies employing one-observation- 
single-act behavior scores may be quite interesting, they are incon- 
clusive, for there is no way of knowing whether attitude would have 
more adequately predicted a related act or the same act on a different 
occasion. Attitudes tovards blacks, for instance, should be related 
to a behavioral pattern manifested in a set of prejudicial actions. 

Finally, in order to examine the relation between attitudes and 
overt behaviors it is necessary to specify v;hat we mean by "overt 
behavior." Certainly it is common to distinguish between pencil and 
paper measures and overt behaviors. However, as Aronson and Carlsmith 
(1968) note, "it is possible to conceive of a continuum ranging from 
behaviors of great importance and consequence for the subject down to 
the most trivial paper and pencil measures about which the subject 
couldn't care less (p. 54)," In practice, researchers have employed 
three distinct forma of behavioral measures: (1) retrospective self- 
reports of behavior (e.g.. Tittle and Hillj 1967), (2) behavioroid 
measures which indicate how & person intends to behave at some point 
in the future (e„g,, DeFleur and Westic, 1958), and (3) actual 
instances of overt behavior (e.g,, LaPiere, 1934), 

Clearly forms (1) and (2) are acceptable measures of behavior 
only to the extent that they are congruent with form (3), actual 
behavior. (This is not to say, of course, that behavioral intentions 
may not be of interest in their own right.) Yet in some circumstances 
retrospective reports may be distorted and intentions unfulfilled. 


As an example of distortions in memory Bern and McConnel (1971) and 
Ross and Shulman (in press) found that subjects who had undergone con- 
siderable attitude change tended to greatly underestimate the amount 
of attitude change that had taken place. Also, Oskamp (1972) notes 
a tendency for people to report that they voted for the winning 
candidate in an election. In 1964, 667o of a group of respondents 
indicated that they had voted for Kennedy in 1970, though only 50% 
actually had done so. With regard to intentions, Linn (1965) found 
that 187o of a group of subjects who agreed to pose for a photograph 
with a Negro failed later to keep their appointments. Likewise, 
Fishbein (1966) obtained a correlation of only .39 between males' 
intentions to engage in premarital sex during a school semester and 
their subsequent retrospective reports, suggesting that males were not 
always able to convert their intentions into behavior. Thus studies 
employing retrospective or behavioroid measures may introduce addi- 
tional sources of variance. The strongest evidence about behavioral 
prediction is derived from research in which samples of actual be- 
havior are obtained or where special attention is given to the 
adequacy of surrogate variables such as self-reports or intentions. 

Testing and Observational Research 

One approach to studying the attitude-behavior rcliitionship is 
to construct tests for both attitude and behavior and then to corre- 
late the results. Vfhile such covariation clearly does not imply 
causality, it may provide evidence that the two concepts are related. 
Several studies have compared attitude scales on the basis of how 

well they predict behavior. These studies have yielded mixed results, 
probably due to variation ±n the quality of the attitude scales 
evaluated (e.g., Kamenetsky, Burgess, and Rowan, 1956; Poppleton and 
Pilkington, 1963; Carr and Roberts, 1965), One study, however, is of 
particular interest, and illustrates the potential of this approach. 
Tittle and Hill (1967) obtained attitude measures toward participation 
in student political activities using Thur stone, Likert, Guttman, and 
semantic differential scaling techniques. Moreover, they used scaling 
techniques to construct behavior scores from self-reports of partici- 
pation in various types of student political activity. The results 
were encouraging: "it is clear that attitude measurement alone ... is 
not totally adequate as a predictor of behavior. However, when it is 
possible to obtain an average association of 0.543 [the gamma statis- 
tic] using a Likert scale in its crude form, it seems entirely possible 
that technical refinements and additional methodological considerations 
could increase predictive efficiency (1967, p. 210-211)," This would 
certainly seem a. fruitful avenue for further research. 

Most of the research in this area has concentrated not on measures 
of attitudes but on the actual observation of a specific behavior, 
li"/hile this observational research is .often socially relevant, it is 
weak methodologically, with single acts being studied far more often 
than patterns of behavior. It is this research that has most fre- 
quently failed to obtain a relationship between attitudes and behavior. 

Much of the observational research has attempted to relate atti- 
tudes as expressed on simple questionnaires to overt behaviors toward 
minority groups, LaPiere's work was an early example of inconsistency 


between such verbal measures and overt behavior, and other similar 
findings have been reported (Kutner, Wilkins, and Yarrow, 1952; 
Saeger and Gilbert, 1950) » A well known study by DeFleur and Westie 
(1958) revealed a greater degree of attitude- behavior consistency. 
A Likert-type measure of attitudes toward blacks was administered to 
250 college students. Subjects in the upper and lower quartiles of 
the attitude distribution were then compared in terms of a simulated 
racial interaction. Each subject was asked if he would pose at some 
time in the future for a photograph with a black of the opposite sex. 
If a subject agreed, he was requested to permit one of several possible 
uses of the photograph, ranging from limited laboratory exposure to use 
in a nationwide integration campaign. DeFleur and Westie found that 
subjects who had reported prejudicial attitudes tended to be less 
willing to have their picture taken and widely distributed. Yet over 
257o of the subjects behaved inconsistently with their attitudes, a 
very high proportion given that only subjects with extremely positive 
or negative attitudes were tested. 

Linn (1965) reasoned that a more specific measure of attitude 
would increase the consistency between attitudes and behavior in the 
DeFleur and Westie situation. Accordingly, Linn conducted a study 
in which the attitude questionnaire items were the same as the be- 
havioral alternatives. The difference between the attitude items and 
the behavioral scale administered four weeks later was that the 
attitude statements were presented as hypothetical commitments as 
opposed to actual cooanitments for the behavioral scale. Despite the 
similarity of the two measures ^ subjects' responses on the two scales 


were not significantly related. Although there are several possible 
explanations for this negative result, not the least of which is the 
possible transparency of the procedure, it is clear that the study is 
more relevant to the relationship between verbally expressed prior 
intentions and behavior rather than the relationship between attitudes 
and behavior, A more descriptive overview of much of this research 
on racial attitudes and behavior is given by Katz (1970, p. 80-90). 

Turning briefly toother content areas, we find striking examples 
of the possible failure of attitudes to predict behavior. Corey (1937) 
related attitudes toward honesty in the classroom with frequency of 
cheating on tests. A Likert-type measure of attitudes failed to pre- 
diet cheating though the students' performance on the tests did. In 
organizational psychology, an area of very active interest has been 
the relationship between job attitudes (usually operationalized as 
reported job satisfaction) and job performance. Vroom (1964) reviewed 
twenty studies relating job attitudes to job performance and found the 
correlations to be disappoint 3.ngly small and often nonsignificant. 

Although these observational studies would in general lead to 
pessimism about a strong attitude- behavior relationship, their 
methodologies are typically too weak to allow a definitive answer 
to the problem. Nevertheless, there is reason to believe that even 
if the studies were better designed high attitude-behavior correla- 
tions would not necessarily have been obtained. For this reason, we 
will concentrate on the general theoretical issues involved rather 
than a study- by-study critique. These issues have taken the form of 
attempts to explain why the attitude-behavior link may not occur. 


The specificity of behavior explanation 

In many instances, the behavior observed may be so specific as to 
have had little relation to the more general attitude measured. 
Fishbein (Ajzen and Fishbein, in press; Fishbein, 1967a) has contended that 
attitudes toward the particular act to be engaged in, rather than 
general attitudes, should be related to behavior. For example, you 
should ask a person specifically about his attitude toward posing for 
a photograph with a black if you wish to predict this behavior. 
Favorable or unfavorable attitudes toward blacks in general, as 
assessed by DeFleur and Westie (1958), should be far less predictive 
of the behavior. Wicker and Pomazal (1971) conducted an experiment to 
test this reasoning. Students were asked to volunteer to participate 
in a psychology experiment after their attitudes towards scientific 
research (general), psychological research (less general), and par- 
ticipating as a subject in a psychology experiment (specific) were 
assessed. While all of the correlations were weak, the only signi- 
ficant association was between volunteering and the specific attitude 
(r - .17). 

One problem in comparing attitudes toward the act to be engaged 
in with iftore general attitudes toward an object or issue is that the 
two are frequently highly correlated, A dominant action may be more 
or less inherently associated with an object so that the two are dif- 
ficult to separate. Ajjeen and Fishbein (1969), for instance, obtained 
extremely high correlations between attitudes toward certain activities 
and attitudes toward objects corresponding to those activities, 
Schwartz and Tessler (1972), in a study described more fully later. 


The specificity of behavior explanation 

In many instances, the behavior observed may be so specific as to 
have had little relation to the more general attitude measured. 
Fishbein (Ajzen and Fishbein, in press; Fishbein, 1967a) has contended that 
attitudes toward the particular act to be engaged in, rather than 
general attitudes, should be related to behavior. For example, you 
should ask a person specifically about his attitude toward posing for 
a photograph with a black if you wish to predict this behavior. 
Favorable or unfavorable attitudes toward blacks in general, as 
assessed by DeFleur and Westie (1958), should be far less predictive 
of the behavior. Wicker and Pomazal (1971) conducted an experiment to 
test this reasoning. Students were asked to volunteer to participate 
in a psychology experiment after their attitudes towards scientific 
research (general) , psychological research (less general) , and par- 
ticipating as a subject in a psychology experiment (specific) were 
assessed. While all of the correlations were weak, the only signi- 
ficant association was between volunteering and the specific attitude 
(r « .17). 

One problem in comparing attitudes toward the act to be engaged 
in with more general attitudes toward an object or issue is that the 
two are frequently highly correlated, A dominant action may be more 
or less inherently associated with an object so that the two are dif- 
ficult to separate, Ajzen and Fishbein (1969), for instance, obtained 
extremely high correlations between attitudes toward certain activities 
and attitudes toward objects corresponding to those activities, 
Schwartz and Tessler (1972), in a study described more fully later. 


used an issue of sufficient complexity as to suggest many behaviors, 
obtaining an average correlation of only .20 between the two attitudes. 
Accordingly, their study provides an interesting test of the specificity 
argument. The results indicated that attitudes toward acts were much 
better predictors of behavioral intentions than attitudes toward 
general issues. Intriguingly, however, in this study attitudes toward 
general issues were also significantly correlated with intentions and 
this association was not washed out by partialing out the effects of 
attitudes toward acts. It would thus appear that both specific and 
general attitudes may predict behavior, though the relationship with 
specific attitudes toward acts may be stronger. 

The behavioral threshold explanation 

In an attempt to reconcile any apparent inconsistency between 
various behavioral manifestations of an attitude, Campbell (1963) 
pointed out that different behaviors have different thresholds of 
appearance. Low threshold indicants of attitudes will occur even with 
a weak attitude whereas high threshold indicants may require a strong 
attitude before they will occur. Campbell noted that LaPiere assessed 
two behaviors with different thresholds: refusing service by letter 
is easier (low threshold) than refusing service in person (high 
threshold). Thus, Campbell concluded that LaPicre's results were not 
inconsistent. By their willingness to perform only low threshold 
behavior, the hotel clerks indicated that they were somewhat, but 
not strongly, anti-Chinese. Rosen and Komorita (1971), moreover, have 
demonstrated that a conventional attitude assessment predicted a 


behavioroid measure significantly less well than an index based in 
part on subjects' intentions to perforin a series of acts varying in 
behavioral thresholds. 

Campbell's formulation has received direct support from a study 
of attitudes toward organ transplantation (Goodmonson and Glaudin, 
1971). The behaviors involved ranged from participating in a tele- 
phone interview (low threshold) to actually signing a legal form 
authorizing posthumous donation of organs (high threshold). The 
results indicated a significant correlation (r - .58) between the 
strength of previously assessed attitudes toward organ transplantation 
and the extremity of the behavior engaged in. Higher threshold behaviors 
tended to be elicited only when the respondents possessed extremely 
positive attitudes. 

Unfortunately the behavioral threshold explanation has been 
associated with the view that attitude-behavior inconsistency can 
never be said to exist (e.g., Sechrest, 1969, p. 147): Inconsistency 
only appears to exist because of the varying thresholds of the be- 
havioral indicants of attitude. As argued earlier, however, not all 
behaviors constitute evidence for a specific attitude. While there 
can be no inconsistency between an attitude and the behaviors that 
operationali^e that attitude, other behaviors a person performs may 
be inconsistent with that behavioral operationalization. Thus the 
behavioral threshold approach confuses the two roles of behavior in 
psychology. It assumes that all behavior relevant to an attitude repre- 
sents an operationalization of that attitude. 


The behavioral threshold perspective is valuable, however, in 
alerting researchers to the problems inherent in measuring two variables 
which may be differentially susceptible to situational cues. Attitudes 
may be frequently less situationally constrained than behavior. As 
pointed out by Hyman (1959), the typical testing setting does not 
involve the forces of everyday life. Fendrich (1967), for example, 
compared attitudes toward blacks as predictors of actual involvement 
in a campus chapter of the NAACP for two groups of subjects whose 
definition of the situation was varied experimentally. The results 
Indicated that subjects encouraged to view the situation as one where 
their attitudes reflected true commitment displayed attitude- behavior 
consistency whereas those encouraged to define the situation as the 
typical "play- like" experiment did not. 

Other variables explanation 

An obvious and frequently proposed explanation of attitude- 
behavior inconsistency holds that while an attitude may affect behavior, 
it is not the sole determinant. Variables other than attitude must be 
taken into account if accurate behavioral prediction is to be achieved. 
For example, Wicker (1971) attempted to predict frequency of church 
attendance by measuring three variables in addition to attitude toward 
the church: perceived con?5cqaence3 of church attendance, evaluation of 
church attendance, and the judged influence of extraneous events on 
church attendance (e.g., the likelihood that the presence of weekend 
guests would affect attendance). The correlation found between 
attitude and frequency of church attendance (obtained from official 


* , 


church attendance records) was .31. The multiple correlation which 
combined the additional three measures with attitude was .50, indi- 
cating an improved behavioral prediction when the other variables were 

Several researchers have attempted more systematic formulations 
of the other variables explanation. Rokeach and Kliejunas (1972) 
have argued that a person *s behavior is determined by two interacting 
attitudes-- his attitude toward the object and his attitude toward 
the situation. To test this hypothesis j Rokeach and Kliejunas assessed 
students' attitudes toward the professors teaching their courses 
(attitude toward the object) and their attitudes toward attending 
classes in general (attitude toward the situation). The average of 
the two attitude measures weighted for their perceived importance was 
found to be a significantly better predictor of self-reports of class 
attendance (r » -,61) than either attitude measure alone (attitude 
toward professor: r « -.20; attitude toward class attendance in 
general: r =» -.46). Fishbein (1967a) has proposed a model in which 
behavior is a function of attitudes toward the behavior and normative 
beliefs regarding the behavior, Similarily Triandis (1971, p. 16) 
would add three other variables to attitude: social norms, habits, 
and expectancies about reinforcement. Sugar (reported in Triandis, 
1971) demoristrated that accuracy of predicting the acceptance of a 
cigarette increased if norms and habits were considered in addition to 
attitudes toward smoking, 

Warner and DeFleur (1969) have taken a somewhat different tack. 
They hypothesized that two situational factors, social constraints and 


social distance, affect whether or not a person's behavior will reflect 
a pro- or anti-black attitude. A field experiment was conducted to 
test this hypothesis. After obtaining a measure of general attitudes 
toward blacks, the investigators sent letters to high and low preju- 
diced subjects asking them to engage in a specific behavior toward 
blacks. The behaviors requested varied in social distance from dating 
a black (low social distance) to contributing to a black charity (high 
social distance). Social constraint was manipulated by having the 
respondent either believe his reply would be kept anonymous (low 
social constraint) or be disclosed toothers via the campus paper (high 
social constraint). Each person received one version of the request 
and was asked to return the letter indicating his agreement or disa- 
greement. The results indicated that with high social constraint, low 
prejudiced subjects acted consistently with their attitudes (agreed 
with the request) when the behavior maintained social distance. On 
the other hand, high prejudiced subjects acted consistently (refused 
the request) if the behavior rediiced social distance. As less than 
2S7o of the subjects answered the letter, however, the results must 
be interpreted with caution. 

Along the same lines. Acock and DeFleur (1972) have formulated a 
"pivotal hypothesis" arguing that "attitude(3) may provide a base- 
line factor for decision-making about action toward the relevant issue 
or object. Against this base-line the individual raises other con- 
siderations, such as the views held by his reference groups, considering 
in particular, possible sanctions for acting one way or another; 
then he makes his action decision (p. 725)." As preliminary support 


for the hypothesis Acock and DeFleur (1972) report a study of atti- 
tudes toward the legalization of marijuana. The best prediction of 
whether subjects would vote in an experimental situation to legalize 
marijuana was obtained only when their attitudes were combined with 
measures of the perceptions of their peers and family in a configura- 
tional approach. 

A recent study by Weitz (1972) indicates just how complex the 
effects of other variables can be. White subjects' verbally expressed 
racial attitudes ("How friendly would you feel toward this person 
in a year's time?") were found to correlate negatively with such 
behaviors as task selection, where the tasks differed in the closeness 
of the interaction with a black and the amount of time required to 
work with the black. Thus subjects displaying more attitudinal 
tolerance tended to behave in a more prejudiced manner. One hypo- 
thesis explains such findings in terms of a basic psychological 
ambivalence in which positive feelings are channeled verbally (and are 
likely to be expressed as attitude 5) while negative feelings are 
channeled into actual behavior (Katz, 1970). Another interpretation is, 
however, that attitudes may in fact be tolerant but other variables 
such as social norms still structure behavior along prejudicial lines. 
These norms essentially cause the positive attitudinal affect to be 
repressed, and can even produce a negative correlation between atti- 
tudes and behavior. As Weitz (1972) points out, on a societal level 
this system in which positive attitudes are not able to overcome dis- 
criminatory racial patterns has been called by Myrdal (1944) the 
"American dilemma." 


Research Based on Models of Attitudinal Organization 

Traditionally, attitudes are thought of as unidimensional, as a 
favorable or unfavorable evaluation or affect. Yet a number of investi- 
gators have treated attitudes as possessing an underlying organization 
or structure. Two versions of this putative organization have been 
closely related to the attitude- behavior question, the cognitive- 
affect ive-conative model and tlie expectancy-value model. The cogni- 
tive-af fective-conative model (Roseberg and Hovland, 1960) structures 
attitudes in terms of three components: the cognitive component is the 
rational, informational basis of attitude; the affective component is 
the feeling of liking or disliking for the attitude object, and the 
conative component is the strength of a person's behavioral tendencies 
toward the attitude object. The major alternative to this model has 
been the expectancy-value model which structures attitudes in terms of 
the beliefs which make up the attitude. These beliefs may be treated 
in different ways but they always refer to the attributes of the 
attitude object which the person considers. Typically each belief 
18 associated with two numerical indices, one gives the probability 
of its occurrence (expectancy) and the other an evaluation of its 
worth (value) . Expectancy-value models generally focus on how these 
two indices combine with c^ach other and across various beliefs in 
order to determine attitude. 

The c ognitive - affect ive-conative approach 

Proponents of a three component model of attitude must first 
demonstrate that there are, in fact, distinct components. If a 



separate conative component of attitude could be identified, we might 
expect it to predict behavior more accurately than the overall atti- 
tude. On the other hand, the three components may only appear to 
differ because they are measured in different ways, Woodmansee and 
Cook (1967) factor analyzed responses to a pool of opinion statements 
concerning racial attitudes. The factors which emerged concerned 
specific content areas such as private rights and could not be 
labeled cognitive, affective, or conative. However, the opinion 
statements used in this study were probably too homogeneous to afford 
an adequate test of the three component model. 

Two studies have employed Canrpbell and Ftske's (1959) multitrait- 
multimethod matrix technique to determine whether there are separate 
components of attitude. Ostrom (1969) constructed four independent 
verbal measures of the cognitive, affective and conative components of 
students' attitudes toward the church. The four methods used were 
Thurstone's equal appearing intervals, Likert's summated ratings, 
Guttraan's scaiogram analysis, and a simple rating scale, Kothandapani 
(1971a, i971b) employed the same four methodologies to measure the 
cognitive, affective, and conative components of attitudes toward 
birth control. The multitrait-multimethod analysis in both studies 
indicated that the three hypothetical components of attitude are 
distinct relative to method variance. These two studies also inves- 
tigated the accuracy with which the three components predicted be- 
havior. Ostrom found that the conative component was generally a 
better predictor of church related behaviors (self report and behavioroid 
measures) than the cognitive and affective components, but that the 


magnitude of this difference was extremely small , Using a stepwise 
discriminant analysis, Kothandapani showed that the conative component 
was the most accurate predictor ot contraceptive behavior (as deter- 
mined from self-reports) and that prediction was not improved by adding 
either the affective or cognitive component, or both, to the prediction 

In view of this research, it appears that the conative component 
of attitude should be further explored. Some earlier research by 
Triandis (1964) is especially relevant. This work investigated the 
factor analytic structure underlying semantic differential type ratings 
of statements regarding various behaviors, Triandis argued that by 
using specific factors underlying the conative component, more accurate 
prediction of behavior may be obtained. 

Although research on the conative component seems promising, it 
is possible to raise a theoretical objection. Perhaps the conative 
component is not really a measure of attitude. That is, the conative 
component may vreli be different from cognitive and affective components 
as measured by various methods, but there is no theoretical basis for 
concluding that it is a separate component of attitude. Rather the 
conative component may provide evidence about general behavioral 
tendencies or intentions. The conative component may thus predict overt 
acts because both the acts aud the conative component are indicants of 
the same concept, behavior. This argument implies that some theoreti- 
cal rationale must be developed for postulating and measuring conation 
as a component of attitude. Until such a perspective is developed, the 
most parsimonious explanation of the conative component is not as a 


basis of attitude but as a measure of behavior. This problem is simi- 
lar to the issues raised in connection with using observations of 
behaviors to infer attitudes. 

Insko and Schopler (1967) have proposed a cognitive-affective- 
conative model in which conation is not employed as a component of 
attitude. In this model, attitudes are defined strictly in terms of 
positive or negative affect; cognitions are beliefs about the rela- 
tionship between objects of affective significance; conation is 
identified as goal directed activity which may be positively or nega- 
tively evaluated. Insko and Schopler hypothesized that people try to 
maintain consistency between attitudes, cognitions and behavior. Thus 
inconsistency between attitudes and behavior should be resolved by 
either a change in attitudes or a change in behavior. Although this 
prediction has not been adequately tested in the context of their 
theory^ Insko and Schopler 's work does provide an example of how 
affect, cognition, and conation can be related by a specific psychologi- 
cal mechanism. 

T he exp e c t an cy- valu e appr o ach 

While several investigators have developed expectancy-value 
toodels of attitude organization (e.g., Peak, 1955, 1958; Rosenberg, 
1956, i960a, 1960b), the work of Fishbein (1963, 1965, 1967a, 1967b, 
in press) is the most relevant to the attitude-behavior relationship. 
Fishbein 's expectancy-value model is given by the equation 


where A is the attitude toward some action or object, B is the strength 
of the belief i about the object or action, and a is the evaluative 
aspect of belief i. This basic moael of attitude organization has 
been extended to the predi-ction of behavior as follows. Fishbein 
believes that behavioral prediction can be increased by employing 
specific attitudes and other variables in addition to attitude. The 

extended model is written as 

B « BI . [A^^^]w^ + [r(NB.)(M^.)lw^ , [2] 

where B is some overt behavior, BI is the intention to perform that 

behavior, A is the attitude toward performing the behavior, NB 

Is the strength of the normative belief j about what other people 

think the individual should do, M is the individual's motivation to 

* c 

comply with normative belief j, and w and w- are empirically derived 
regression weights. Note that Equation [1"] can be substituted into 
Equation [2], An interesting feature of the model is that overt 
behavior and behavioral intentions are seen as being approximately 
the same, Fishbein (in press) argues that most behavior is under 
volitional control and that intentions will be very closely reflected 
in behavior if they are measured properly, i.e., temporally close to 
the behavior, etc. Equation [2] then predicts these behavioral 
intentions from the additive combination of an individual's attitude 
toward the action and his perception of and susceptibility to norma- 
tive pressures regarding the behavior. Behavioral prediction thus 
rests on an expectancy-value model of both attitude and normative 


Several studies have tested Fishbein's attitude-behavior model. 
One of these provides a useful illustration. Fishbein (1966) attempted 
to predict the occurrence of prema..ital sexual intercourse for male and 
female subjects. Behavioral intentions were correlated with actual 
behavior though the association was higher for females than for males 
(r » .69 versus r = ,39). The multiple correlations between behavior 
and the attitudinal and the normative components of the model were 
quite high for both males and females (r = .89 and r = ,94 respectively) 
The regression weights, however, differed for males and females. For 
females, the attitudinal component received the most weight in the 
regression equation while, for males, the normative component contri- 
buted more. This finding is also reflected in a higher correlation 
between attitudes and sexual behavior for females (r » .92) than 
males (r » .52). Again we see the complicated interaction of attitu- 
dinal and situational factors. 

Ajzen and Fishbein (in press) review nine studies that support 
the model. All of these studies obtained relatively high correlations 
between specific attitude measures (A ) and behaviors such as inten- 
tions to engage in recreational activities (Ajzen and Fishbein, 1969) 
or cooperative intentions and choices in a Prisoner's Dilemma game 
(Ajaen and Fishbein,, 1970)' Schwartz and Tessler (1972) report an 
excellent study exploring the adequacy of Fishbein's model in predic- 
ting intentions about six kinds of medical transplant donations. The 
tested version of Fishbein's model consisted of three components, 
attitude toward the act, social normative beliefs, and personal norma- 
tive beliefs in the sense of moral obligation. Although previous work 
(Ajzen, 1971; Ajzen and Fishbein, 1969, 1970, in press) has suggested 


omitting personal normative beliefs and the motivation to comply from 
the model, Schwartz and Tessler argue that the former may prove valuable 
if operationalized as moral obliges :ion. The results of the study re- 
vealed significant correlations between all three components and 
behavioral intentions for all six transplant donations. An average 
of more than 507c of the variance in intentions was explained. This 
proportion was not even reduced very much for a crossvalidation sample 
using the original regression weights. A step-wise regression analysis 
indicated that all three components contributed significantly to the 
explained variance, 

Schwartz and Tessler also examined whether Fishbein's model media- 
ted the infliience of six exogenous variables. The relationship of four 
variables, attitude toward the object, age, religiosity, and occupa- 
tional prestige, with intentions was not eliminated by controlling for 
the effects of the model's components. This result along with similar 
findings by Ajzen and Fishbein (1969, 1970) casts doubt on the suf- 
ficiency of Fishbein's model. Finally, Schwartz and Tessler obtained 
measures ox volunteering behavior too, but found that only personal 
normative beliefs predicted actual behavior. The authors conclude 
that Fishbein's model may be a better predictor of intentions than 
behavior. While this is probably true, these results really serve to 
highlight the myriad of other variables besides attitude which affect 
behavior and to a lesser extent intentions. 

Expectancy-value models have also been popular in the area of 
crganiziational behavior. Vroom (1964) has devised an expectancy- 
value model for job performance. Empirical studies, however, have 
generated only weak support for these models (e.g., Galbraith and 


Coranings, 1967; Hackman and Porter, 1968; Lawler, 1968; Lawler and 
Porter, 1967). Graen (1969) has provided some possible limitations and 
extensions of Vroom's model. 

Attitude Change- Behavior Change Research 

Most of the studies of attitudes over the past two decades have 
focused on measuring attitude change as a result of a specific experi- 
mental manipulation. Very few of these studies have also included a 
measure of behavior change. In an early review of this area, Festinger 
(1964) was able to find only three studies that incorporated measures 
of both attitude and behavior (Fleishman, Harris and Burtt, 1955; 
Janls and Feshback, 1963; MacCoby, Romney, Adams, and MacCoby, 1962). 
While all three studies successfully modified attitudes, corresponding 
shifts in behavior were not obtained, 

A number of more recent investigations have been specifically 
designed to test whether behavior change accompanies attitude change. 
In two studies, Greenwald (1965, 1966) found that a communication 
changed both children's attitudes toward a task and their performance 
on the task. Freednan (1965) failed to produce the expected attitude 
change, yet behavior change was obtained. Hendryk and Seyfrled (1972) 
developed a novel paradigm for studying the consequence of attitude 
change. Experimental and control subjects were yoked on the bases of 
initial attitude responses. Experimental subjects were then exposed 
to a persuasive conmunication and their attitudes were reassessed 
(post-test). The next stage of the experiment capitalized on Byrne's 
research on interpersonal attraction which establishes a link between 
attitude siJDilarlty and liking. Hendryk and Seyfrled showed subjects 



in both conditions the attitude responses of two strangers. The atti- 
tude responses of one stranger were similar to the subjects' pre-test 
attitudes. The attitude responses of the second stranger were identi- 
cal to the experimental subjects' post-test attitudes. Rating of 
attraction showed that the experimental subjects preferred the post- 
test attitude stranger while control subjects preferred the pre-test 
attitude stranger. Though measures of overt behavior were not obtained, 
Hendryk and Seyfried demonstrated that effects of attitude change spread 
to a conceptually related response, liking for an anonymous stranger. 
Further extensions of this type of paradigm could provide important evi- 
dence linking attitude change and behavior change. 

Other research which has simultaneously included measures of atti- 
tude and behavior has yielded inconsistent results (e.g., Weick, 1964; 
Leventhal, 1970). Leventhal reported five studies on fear arousing 
communications that successfully produced attitude change, but only two 
of these studies also yielded changes in behavior. In summary, changes 
in attitude do not always appear to produce changes in behavior, A 
number of factors are. likely to affect whether or not attitudes and 
behavior will covary and some of the most important of these are dis- 
cussed below. 

The fu nct iona l ricitura of the att itude-brhavlor relationship 

Even if Hshavior is functionally related to attitudes, it is not 
necessary that any change in attitude result in a change in behavior. 
The precise nature of this relationship might take a number of forms 
as illustrated in Figure 1. For the first curve (1), large changes in 
attitude (ac) produce only small changes in behavior (be). If 


attitudes were related to behavior by this function, it would not be 
surprising if research successfully obtained attitude change but was 
not able to detect the resulting small amount of behavior change. The 
second curve (2) depicts a more complicated functional relationship. 
Here the slope changes with the region of the curve examined. Different 
portions of the attitude scale yield substantial differences in the 
amount of behavior change. A moderately religious person, for example, 
who becomes less religious may change his behavior far less than an 
extremely religious person who loses his former zeal. 

Insert Figure 1 about here 

Attitude change may be unstable 

Festinger (1964) offered a further explanation for the frequent 
lack of correspondence between attitude change and behavior change. 
Attitude change may be transitory and unstable unless it is supported 
by accompanying environmental changes. Such a hypothesis is illus- 
trated by Newcomb's (1943, 1963, 1967) finding that girls at Benning- 
ton College who came into the more liberal academic community tended 
to adopt more liberal attitudes. Newcomb relates the subsequent per- 
sistence of these attitudes over a twenty- five year period to the 
environmental support provided by the girls' husbands who also 
possessed liberal views. In the absence of a supportive informational 
environment, attitudes may not persist long enough to affect behavior. 



Greenwald (1966), in a study referred to earlier, found that com- 
mitment to a behavior made it more resistant to change. Whereas both 
the behavior and attitudes of noncommitted subjects were influenced by 
a communication, subjects conimitted to a conflicting behavior changed 
their attitudes toward a task but not their performance. These results 
suggest that a prior commitment inay lock a person into a behavior and 
decrease its susceptibility to any change in attitude. For example, 
once a couple formally announces their engagement, it may become much 
more difficult for any subsequent negative attitude change to affect 
their behavior. 

Behavior is sltuatlonally constrained 

While attitude change is generally covert, behavior change is 
often public. As a result, changes in behavior are more likely than 
changes in attitude to result in negative or positive consequences. 
The reformed bigot may find it very costly to change his behavior in a 
society, such as South Africa, r'lere racial prejudice is the norm. A 
man may become a model prisoner to gain parole, though his attitudes 
toward crime remain unaltered. 

Behavior-Attitude Research 

It is necessary to distinguish between attitudes as causes 
(determinants) of behavior and attitudes as predictors of behavior. 
Few researchers have explicitly addressed this problem. The implicit 
assumption in the literature is that attitudes should predict behavior 
because they are a cause of behavior. This assumption may be false. 


It is possible to argue that attitudes may serve as convenient surro- 
gates for predicting behavior in advance, but though they are associated 
with behavior they are not major determiners of behivior. It is incum- 
bent on such an argument, of course, to provide some explanation for 
how such a non-causal association could arise. The most viable explana- 
tion simply reverses the asrtumed causal direction. Attitudes may not 
cause behavior, rather behavior causes attitudes: attitudes predict 
behavior only because the two are related through previous performance. 
Past behavior molds our attitudes which will in turn predict future 
behavior if this behavior is related to prior performance, 

Bern (1968) has indeed suggested that there is more evidence in 
support of the counterintuitive notion that behavior affects attitudes 
than there is for the common sense assumption that attitudes determine 
behavior. The bulk of this evidence comes from research on cognitive 
dissonance theory (Festinger, 195?) and Bem's own theory of self- 
perception (1965, 1967, 1972), However, just as it is now becoming 
clear that attitudes do not always influence behavior, similarly 
neither dissonance nor self-perception theory postulate that behavior 
always affects attitudes. Behavior that is clearly perceived to be 
under the control of extrinsic factors such as reward or punishment is 
not hypothesised to influence attitudes, and only behavior that the 
Individual perceives himself to be enacting of hiu own free will 
determines attitudes. 

There have probably been ainost as many failures to obtain the 
expected dissonance results of a behavior- attitude link (e.g., Collins, 
Ashmore, Hornbeck, & ¥hitney, 1970; Melson, Calder, & Insko, 1969) as 


there have been failures to demonstrate the attitude-behavior connection, 
In recent years research in both areas has attempted to delineate the 
crucial parameters j to determine when the attitude-behavior or behavior- 
attitude association is likely to occur. With regard to the behavior- 
attitude association. Cooper and Worchel (1970) found that counter- 
attitudinal behavior resulted in attitude change only when the behavior 
was enacted for low inducement and it resulted in negative consequences 
for another person. Collins and Hoyt (1972) extended this finding by 
demonstrating that the individual must feel personally responsible for 
the behavior if attitude change is to result. Finally Calder, Ross, 
and Insko (1972) demonstrated the importance of choice: Negative con- 
sequences and low financial inducement led to attitude change only 
when subjects volunteered to perform the counter attitudinal behavior 
(choice condition). When subjects were forced to perform a behavior 
resulting in negative consequences, high inducement produced more 
attitude change than lovr inducement, (It is likely that choice as 
manipulated by Calder at al. and personal responsibility for conse- 
quences as manipulated by Collins and Hoyt, while operationally dis- 
tinct, are closely related coucnptually. High choice should induce 
perccptioriS of high parsoi^al responsibility end low choice perceptions 
of minimal personal rer^ponaibiii ty) . 

It should also be noted that the behavior-attitude link has been 
observed for behaviors r,het were not originally counterattitudinal 
(e.g., Valine, 1966; Kiesler, Nisbett, and Zanna, 1969; Ross, Insko, 
and Ro3s^ 1971). Again the *iffects of behavior appear to interact 
with other variables though. Kiesler and Sakumura (1966), for 


example, cienons traced that subjects who were paid $5.00 for stating a 
position agreeing with their own point of view were subsequently more 
■•Ailnerable to countercommunicati ms than subjects vho had been paid 
only $1.00« The greater the external reward for a behavior, the less 
it seems to affect one's private attitude. 

In summary, the behavior- attitude association is complicated by 
various interacting factors that determine the precise nature of the 
relationship. A number of recent studies have attempted to systemati- 
cally identify the most important of these factors. In this sense 
attitude-behavior research and behavior-attitude research are pro- 
ceeding along siraiiar lines. However, there is a great need to examine 
the behavior-attitude link as a possible alternative explanation in 
studies relating attitudes to behavior. To evidence for a 
causal relation, such studies must control for the effects of previous 
performance which might have produced an attitude which predicts future 
behavior (if it is similar to the prior behavior) but does not actually 
cause that behavior. Sincf* it seems likely that both causal processes 
are at vjork, it may prove most aifiicuit to disentangle them. 


Full consideration of research relevant to the attitude-behavior 
question has indicated that evidence for the proposition that atti- 
tudes are related to behavior is not as weak as many social scientists 
have contended. In general, the research indicates that attitudes will 
correlate with behavior when: 


1. standard attitude scale techniques and multiple act behavior 
scores are employed (e.g., Tittle and Hill, 1967), 

2. attitudes toward the act and attitudes toward the situation in 
which the act occurs are taken into account (e.g., Fishbein, 
1967a; Rokeach and Kliejunas^ 1972), 

3. the conarive component of attitude is used as a basis of 
prediction (e.g., Kothandapani, 1971a), 

4. situational constraints do not produce behavior that is 
inconsistent with attitudes (e.g., V/arner and DeFleur, 1969; 
Weitz, 1972). 

Definitive statements concerning the relationship between atti- 
tude change and behavior change must be made with greater trepidation 
as only a few studies have been directly concerned with this problem. 
The proposition that attitudes affect behavior does not imply that the 
two arc linearly related. It does follow, however, that attitude 
change must produce behavior change in some instances. More research 
is required to determine the conditions under which attitude change 
yields behavior change. 

A considerable amount of research has focused on the behavior- 
attitude association. At the present time the research on counter- 
attitudinai role playing points to three important interacting varia- 
bles: choice (or personal responsibility), financial inducement, and 
consequences (Calder, Ross 6e Insko, 1973; Collins & Hoyt, 1972), 
identification of these factors has resolved much of the inconsis- 
tency in the literature dealing with attitude change following counter- 
attitudinal behavior. The behavior-attitude link, however, has not 
tested as a possible aitertiative explanation in attitude-behavior studies. 


Allport (1935) noted that attitude has been an indispensible con- 
cept since the very beginnings of experimental psychology. While the 
present review is by no means coaclusive, the data certainly suggest 
that the concept of attitude still has an important role to play. 
After nearly one hundred years it has not yet outlived its usefulness. 



Acock, A. C. and DeFleur, M. L. A conf igurational approach to con- 
tingent consistency in the attitude -behavior relationship. 
American Sociological Review , 1972, j6, 714-726. 

Ajzen, I. Attitudinal vs. normative messages: An investigation of 

the differential effects of persuasive communications on behavior, 
Sociometry , 1971, 34, 263-280. 

Ajzen, I. and Fishbein, M. The prediction of behavioral intentions in 
a choice situation. Journal of Experimental Social Psychology , 

1969, 2, 400-416. 

Ajzen, I. and Fishbein, M. The prediction of behavior from attitudinal 

and normative variables. Journal of Experimental Social Psychology , 

1970, _6, 466-487. 

Ajzen, I. and Fishbein, M. Attitudinal and normative variables as predictors 
of specific behaviors: A review of research generated by a theoretical 
model . Journal of Personality and Social Psychology , in 


Ajzen, I. and Fishbein, M. Attitudes and normative beliefs as factors 
influencing behavioral intentions in hypothetical situations 
involving risk. Journal of Personality and Social Psychology , 
1972, 21,, 1-9. 

Aronson, E. and Carlsmith, J. M. Experimentation in social psychology. 
In G. Lindzey and E. Aronson (Eds.). The handbook of social 
psychology. Vol. 2. Reading, Massachusetts: Addison-Wesley, 1968, 
Pp. 1-79. 


Bern, D. J. An experimental analysis of self-persuasion. Journal of 

Experimental Social Psychology > 1965, ^, 199-218. 
Bern, D. J. Inducing belief in false confessions. Journal of Personality 

and Social Psychology , 1966, 2. 707-710. 
Hem, D. J. Self-perception: An alternative interpretation of cognitive 

dissonance phenomena. Psychological Review , 1967, 2^, 183-200. 
Bern, D. J. Attitudes as self -descriptions: Another look at the 

attitude-behavior link. In A. G. Greenwald, T. C. Brock, and 

T. M. Ostrora (Eds.), Psychological foundations of attitudes . 

Academic Press, 1968, Pp. 197-216. 
Bern, D. J. Self-perception theory. In L. Berkowitz (Ed.), Advances in 

experimental social psychology , Vol. 6. New York: Academic Press, 

1972, Pp. 1-62. 
Bern, D. J. and McConnell, H. K. Testing the self -perception explanation 

of dissonance phenomena: On the salience of premanipulation 

attitudes. Journal of Personality and Social Psychology , 1970, JL4, 

Celder, B. J., Insko^ C. A.^ and Yandell, B. The relation of cognitive 

and memorial processes to persuasion in a simulated jury trial. 

Journal of Applied Social Psychology , in press , 
Calder, E. J., Ross, M., and InskOj C. A. Attitude change and attitude 

attribution: Effects of incentive, choice, and consequences. 

Journa l of Personality and Social Psychology , 1973, 25^, 84-99. 
Campbell, D, T. The indirect assessment of social attitudes. Psychological 

Bulletin , 1950, 47, 15-38. 


Campbell, D. T. Social attitudes and other acquired behavioral 

dispositions. In S. Koch (Ed.)j Psychology : A study of a 

science . Vol. 6. New York: McGraw-Hill, 1963. 
Campbell, D. T. and Friske, D. W. Convergent and discriminant 

validation by the multitrait-multimethod matrix. Psychological 

Bulletin , 1959, 56, 81-105. 
Campbell, D. T., Kruskal, W,, and Wallace, W. Seating aggregation as 

an index of attitude. Sociometry , 1966, 29^, 1-15. 
Carr, L. and Roberts, S. 0. Correlates of civil-rights participation. 

Journal of Social Psychology . 1965, 67, 259-267. 
Collins, B. E. and Hoyt, M. F. Personal responsibility-for-consequences: 

An integration and extension of the "forced compliance" literature. 

Journal of Experimental S ocial Psychology , 1972, 8^, 558-593. 
Collins, B. E., Ashmore, R. D. , Hornbeck, F. W. , and Whitney, R. E. 

Studies in forced compliance: XIII and XV. In search of a 

dissonance-producing forced-compliance paradigm. Representative 

Research in Social P sychology , 1970, _1, 11-23. 
Cook, S. W. and Selltiz, C. A multiple-indicator approach to attitude 

measurentent. Psychological Bulletin , 1964, _62, 36-55. 
Cooper, J. and Worchel, S. Role of undesired consequences in arousing 

cognitive dissonance. Journal of Personality and Social Psychology , 

1970, 16, 199-206. 
Corey, S* M. Profesaed attitudes and actual behavior. The Journal of 

Educational Psychology « 1937, 38, 271-280. 


DePleur, M. L. and Westie, F. R. Verbal attitudes and overt acts: An 

experiment on the salience of attitudes. American Sociological 

Review , 1958, 23, 667-673. 
DeFleur, M. L. and Westie, F. R. Attitude as a scientific concept. 

Social Forces , 1963, 4^, 17-31. 
Deutscher, I. Looking backward: Case studies on the progress of 

methodology in sociological research. The American Sociologist , 

1969, 4, 35-41. 
Doob, L. The behavior of attitudes. Psychological Review , 1947, 54, 

Etzioni, A. Human beings are not very easy to change after all. 

Saturday Review , June 3, 1972, 45-47. 
Fendrich, J. M. A study of the association among verbal attitudes, 

commitment and overt behavior in different experimental situations. 

Social Forces , 1967, 45, 347-362. 
Feetinger, L. A theory of cognitive dissonance. Stanford California: 

Stanford University Press ^ 1957. 
Festtnger, L. Behavioral support for opinion change. Public Opinion 

Quarterly, 1964, 28, 404 --A 17. 
Flshbein^ K. An investigation of the relationships between beliefs about 

an object and the attitudes toward that object. Human Relations , 

1963, 16, 233-240. 
Fishbeln, M. A consideration of beliefs, attitudes and their 

relationships. In I. Steincr and M. Fishbein (Eds.), Current Studies 

in Social Psychology . New York: Holt, Rinehart, & Winston, 1965, 

Pp. 107-120. 



Fishbein, M. Sexual behavior and prapositional control. Paper 
presented at the Psychonomic Society meetings, 1966. 

Fishbein, M. Attitude and the prediction of behavior. In M. Fishbein (Ed.), 
Readings in attitude theory and measurement . New York: Wiley, 
1967a, Pp. 477-492- 

Fishbein, M. A consideration of beliefs and their role in attitude 

measurement- In M. Fishbein (Ed..), Readings in attitude theory 
and measurement . New York: Wiley, 1967b, Pp. 257-266. 

Fishbein, M. The prediction of behavior from attitudinal variables. 

In C. D. Mortensen and K, K. Sereno (Eds.), Advances in communica- 
tions research . New York: Harper, Row and Co., in press. 

Fishbein, M. and Ajzen, I. Attitudes and opinions. In P. H. Mussen 

and M. R. Rosenweig (Eds.), Annual Review of Psychology , Vol. 23. 
Palo Alto: Annus! Review, Inc., 1972, Pp. 487-544. 

Fleishman, E., Harris, E., and Burtt, H. Leadership and supervision 
in industry: An evaluation of a supervisory training program. 
Bureau of Educational Research, Ohio State University, 1955. 

Freedraan, J. L, Long-term behavioral effects of cognitive dissonance. 
Journal of^ E xperitTiental Social Psychology , 1965, J^, 145-155. 

Galbraith, J. ami Cuiwuinga, L. L. An empirical investigation of the 

motivatioiaal determinants of task performance: Interactive effects 
1>etiiit^en instrumental ity-valence and motivation ability. Organizational 
Behavior and Human Performance, 1967, 2, 237-257. 


Goodmonson, C. and Glaudin, V. The relationship of commitment- free 

behavior and commitment behavior: A study of attitude toward 

organ transplantation. Journal of Social Issues , 1971, 27, 171-183. 
Graen, G. B. Instrumentality theory of work motivations: Some experimental 

results and suggested modifications. Journal of Applied Psychology 

Monographs , 1969, 53^, (Whole no. 2, part 2). 
Greenwald, A. G, Behavior change following a persuasive communication. 

Journal of Personality , 1965,2^5 370-391. 
Greenwald, A. G. Effects of prior commitment on behavior change after 

a persuasive communication. Public Opinion Quarterly , 1966, 29 , 

Hackman, J. R. and Porter, L. W. Expectancy theory predictions of work 

effectiveness . Organizational Behavior and Human Performance , 

1968, 2* 417-426. 
Hendrick, G. and Seyfried, B. A. Attitude change and behavior change. 

Unpublished manuscript, 1973. 
Hytnan, H. H. Inconsistencies as a problem of attitude measurement. 

Jou.r-nal of So cial I»8uee , 1959 , 5, 38-42 « 
X.nskOj C. A. and Scboplcir, J. Triadic consistency: A statement of 

aff«ctive»cognitive-conative consistency. Psychological Review . 

1967, 74, 361-376. 
Janis, I. L. and Feshbach, S. Effects of fear-arousing communications. 

Journal of Abnormal and Social Psychology, 1953, 48, 78-92. 


Kamenetsky, J., Burgess, G. G. and Rowan, T. The relative effective- 
ness of four attitude assessment techniques in predicting a 
criterion. Educational and Psychological Measurement , 1956, 16 , 

Katz, I. Experimental studies of negro-white relationships. In 

L. Berkowitz (Ed.), Advances in experimental social psychology . 
Vol. 5. Academic Press, 1970, Pp. 71-117. 

Kiesler, C. A., Collins, B. E. and Miller, N. Attitude change . New York: 
Wiley, 1969. 

Kiesler, C. A., Nisbett, R. E. , and Zanna , M. P. On inferring one's 
beliefs from one's behavior. Journal of Personality and Social 
Psychology , 1969, 4, 321-327. 

Kothandapani, V. Validation of feeling, belief, and intention to 
act as three components of attitude and their contribution to 
prediction of contraceptive behavior. J ournal of Personality and 
So cial Psy chol ogy , 1971fi, 19, 321-333. 

Kothfindapani , V . A^ psychological approach to the prediction of 

contraceptv-ve behavior. Monograph 15, North Carolina Population 
Center, University of North Carolina, 1971b. 

Kutner, B»» Wilkins, C.j and Yarrow, P. R. Verbal attitudes and overt 
behavror involving racial prejudice. Journal of Abnormal and 
Social Psvchology, 1952, 47, 649-652. 

»..riiii»i»i.i .1.1111 1 I iriiii I I lim III ^if ' ' .__«.' 


LaPiere, R. T. Attitudes versus actions. Social Forces , 1934, 13 . 

Lawler, E. E, A correlational-causal analysis of the relationship 

between expectancy attitudes and job performance. Journal of 

Applied Psychology , 1968, ^, 462-468. 
Lawler, E. E. and Porter, L, W. Antecedent attitudes of effective 

managerial performance. Organizational Behavior and Human 

Performance . 1967, 2, 122-142. 
Leventhal, H. Findings and theory in the study of fear communications. 

In L, Berkowitz (Ed.), Advances in experimental social psychology . 

Vol. 5. New York: Academic Press, 1970, Pp. 119-186. 
Lmn, L. S. Verbal attitudes and overt behavior: A study of racial 

discrimination. Social Forces , 1965. 43, 353-364. 
Maccoby, N. , Romney, A. K., Adams, J. S., and Maccoby, E. E. 

Critical p<sriods in seeking and accepting information . Paris-Stanford 

Studies in Cormtiunicetion, Institute for Communication Research, 1962 

(cited in Festiager, 1964). 
McGuire, W. J. The nature of attitudes and attitude change. In G. Lindzey 

find E. Aronsom (Eda.), The handbook of social psychology . Vol. 3. 

fHQi? York: Addison-Wesley, 1968, Pp. 136-314. 
Kelson, W. H. , Galdcr, B. J., and Insko, C. A, The social psychological 

status of reward. Psvchonoroic Science, 1969, 17, 240-242. 


Mischcl, W. Personality and assessment . New York: Wiley, 1968. 
Myrdal, G. An Ame rican dilen-ima. New York: Harper and Row, 1944. 2 vols. 
Newcomb, T. M. Personality and social change . Dryden, 1943. 
Newcomb, T. M. Persistence and regression of changed attitudes: Long 

range studies. Journal of Social Issues ^ 1963, 2J9» 3-14. 
Newcomb, T. M. Persistence and change ; Bennington College and its 

students after 25 years . New York: Wiley, 1967. 
Oskamp, S. Methods of studying human behavior. In L. S. Wrightsman (Ed.), 

Social psychology in the seventies . Belmont, California: Brooks-Cole, 

Ostrom, T. M. The emergence of attitude theory: 1930-1950. In A. G. 

Greenwald, T. C. Brock, and T. M. Ostrom (Eds.), Psychological 

f oundations of a ttitudes . New York: Academic Press, 1968, Pp. 1-32. 
Ostrom, T. M. The relationship between the affective, behavioral, and 

cognitive conipon^fnts of attitude. Journal of Experimental Social 

Psychology, 1969, 5^, 12-30. 
Peak, H. Attitude and ffiotivauion. In M. R. Jones (Ed.), Nebraska 

Symg>oj5lum on Motiv ation. Vol, 3. University of Nebraska Press, 

1955, Pp.. 149-189, 
Peak, H. Psychological structure and psychological activity. 

Fsyohalo^ical Review, 1953, _65, 325-347. 
Poppletcn, ?. K. and Pilkington, G. W. A comparison of four methods 

of scoring an attitude scale in relation to its reliability and 

validity. British Journal o_f Social and Clinical Psychology , 1963 , 

3, 36-39. 


Kokeach, M. and Kliejoaas, ?. Behavior as a function of attitude-toward- 

object and attitude-toward-situatlon. Journal of Personality, 

«Qfi. i£Ei^ i§.X2i2£l2SX> ^^^2' li' 194-201. 
Kosen, B. and Komorita, S. S. Attitudes and action: The effects of 

behavioral intent and perceived effectiveness of acts. Journal ^ 
Personality , 1971, 39, 189-203. 

Rosenberg, M. J. Cognitive structure and attitudinal affect. Journal of 

Abnormal and Social Psychology , 1956, ^, 367-372. 
Rosenberg, M. J. Cognitive reorganization in response to the hypnotic 

reversal of attitudinal affect. Journal of Personality , 1960a, 28 , 

Rosenberg, M. J. An analysis of affective-cognitive consistency. In 

M. J. Rosenberg, C. I. Hovland, W. J. McGuire, R. P. Abelson, and 

J. W. Brehra (Eds.), Attitude organization and change . New Haven: 

Yale University Press, 1960b, Pp. 15-64, 
Rosenberg, M. J. and Hovland, C. I. Cognitive affective and behavioral 

components of attitude. In M. J. Rosenberg, C. I. Hovland, 

M. W. McGwire, R. P, Abelson, and J. W. Brehm (Eds.), Attitude 

Organization and change . New Kaven: Yale University Press, 1960, 

Pp. I -(A. 
Ross, M. and Shulmanj, R. S. Increasing the salience of initial attitudes: 

Dissonance versus self-perception theory. Journal of Personality 

£ikI Sociift 1_ Psychology , in press . 
Ross, M., Insko, C. A., and Ross, H.. S. Self-attribution of attitude. 

JQ^-^'mIv of. Feraonality and Social Psychology , 1971, JL7, 292-297 . 


Saenger, G. and Gilbert, E. Customer reactions to the integration of 

Negro sales personnel. Interna tional Journal of Opinion and 

Attitude Research , 1950, 4_, 57-76. 
Schwartz, S. H. and Teseler, R. C. A test of a model for reducing 

measured attitude-behavior discrepancies. Journal of Personality 

and Social Psycholog y, 1972, 24, 225-236. 
Sechrest, L. Nonreactive assessment of attitudes. In E. P. Willems 

and M. L. Rausch (Eds.)* Naturalistic viewpoints in psychological 

research . New York: Holt, Rinehart, and Winston, 1969. 
Skinner, B. F. Beyond freedom and dignity . New York: Knoff, 1971. 
Thurstone, L. L. The measurement of social attitudes. Journal of 

Abnormal and Social Psychology , 1931, 26, 249-269. 
Tittle, C. R. and Hill. R. J. Attitude measurement and prediction of 

behavior: An evaluation of conditions and measurement techniques. 

Sociometry, 1967, 30, 199-213. 
Triandis, H. C, Exploratory factor anlyses of the behavioral components 

of social attitudes. Jour nal of Abnormal an d Social Psychology , 

1964, 68, 420-430. 
TrlandlSj, H. C. Attitude and attitude change . New York: Wiley, 

Valins, S. Cogiii?:ive effects of ifalse heart-rate feedback. Journal 

£i. JP€''"^sonality and S ocial Psychology , 1966, 4^, 400-408. 
Vroom, V. H. Work and motivation. New York: Wiley, 1964. 


Warner, L. G. and DeFleur, M. L. Attitude as an interactional concept: 

Social constraint and social distance as intervening variables 

between attitudes and action. American Sociological Review , 

1969, 34, 153-169. 
Webb, E. J., Campbell, D. T., Schv/artz, R. B., and Sechrest, L. 

Unobtrusive measures ; Nonreactive research in the social sciences . 

Chicago: Rand McNally, 1966. 
Weick, K. E. Reduction of cognitive dissonance through task enhancement 

and effort expenditure . Journal of Abnormal and Social Psychology . 

1964, 68, 533-539. 
Wicker, A. W. Attitudes versus actions: The relationship of verbal 

and overt behavioral responses to attitude objects. Journal of 

Social Issues , 1969, 25, 41-78, 
Wicker, A. W, An examination of the "other variables" explanation of 

attitude-behavior inconsistency. Journal of Personality and 

Soci al Psychology , 1971, 1% 18-30. 
Wicker, A. W. and Pomasal, B.. J. The relationship between attitudes 

and behavior as & function of specificity of attitude object 

and presence of a significant person during assessment conditions. 

Represep.tative Re search in Social Psychology , 1971, 2^, 26-31 . 
Weitx, S. Attitude, voice, and hehavior: A repressed affect model of inter- 
racial interaction. Journa l of Personality and Social Psychology , 

1972, 24, 14-21. 

Figure Caption 

Fig. 1 Hypothetical functional relationships between attitude and 

Note. -"The dotted lines indicate changes in behavior as a 
function of changes in attitude. Notice that the 
three ac lines are of equal magnitude. 









College of Commerce and Business Administration 
University of Illinois at Urbana-Champaign 

May 4, 1973 

Bobby J. Calder, University of Illinois 
Michael Ross, University of Waterloo 


::;j .